Google's Gemini 4 Carbon ranks #1 on Artificial Analysis Intelligence Index at 59, opens first to cyber defenders

Google announced Gemini 4 Carbon on October 28, 2026, a model it calls its strongest, with state-of-the-art cybersecurity and long-horizon capabilities. According to Google, Carbon ranks #1 on the recently updated v5.0 of the Artificial Analysis Intelligence Index with a score of 59. Koray Kavukcuoglu, SVP at Google DeepMind and Chief AI Architect at Google, wrote the announcement.

Access and pricing

Carbon is not yet generally available. Starting this week, Google will make it available to a subset of cyber defenders, then phase in additional trusted partners through its Fairwind Program. Google says it is gathering feedback and iterating on guardrails before releasing Carbon to developers, businesses and consumers through its product surfaces "as soon as possible." It also says it is engaged in the U.S. government's voluntary process for pre-release model testing.

Carbon will launch at the same introductory price as Gemini 4 Argon, which Google says launched two weeks ago:

• Input: $2 per 1M tokens
• Output: $10 per 1M tokens
• Cached input: priced at 95% off the input token price

Cybersecurity benchmarks

Google reports the following external results, which it says were evaluated without guardrails:

• SEC-bench Pro (finding vulnerabilities): 88.2%
• ExploitGym (exploitation): 54.8%, which Google describes as a new state of the art
• CWE-bench from Collinear (patching): 71.0%

On internal benchmarks, Google says Carbon delivers more than 2x uplift in success rate for generating secure code over Argon. It also says the Chrome team reported Carbon performed more than 10X better than another lab's large frontier model when generating high-quality patches. Wiz, which Google calls its leading cloud security platform, observes Carbon leading on benchmarks for web penetration testing.

Coding and agentic performance

Google says Carbon is built for sustained agentic workflows and that internal teams have seen it run for days on multi-step engineering tasks, such as root-causing and patching subtle bugs across large codebases and low-level systems. On WebDev Arena, Carbon gains 89 Elo points to land at #4 with a score of 1769. Google also cites improvements in scientific tasks, full-stack web development and multimodal understanding.

Safety and alignment claims

Carbon uses the same safety and security stack as Argon: model training, real-time misuse defenses, prompt-injection resilience and chain-of-thought monitoring. Google's reported figures, all from its own testing unless noted:

• Successful cheating on Terminal-Bench 4.0 fell to 0.0%, with cheating attempts at 0.4%, while overall accuracy improved.
• High-severity misalignment alerts in live coding environments with internal teams dropped by almost 80%.
• Unsupported claims fell by approximately 2x versus Argon.
• Overreach fell 26% and under-service fell 29%.
• On Gray Swan's Indirect Prompt Injection benchmark, Carbon's attack success rate is 0.0%, versus 0.7% for Argon.

Google also says Carbon retrieves relevant sources and surfaces discrepancies to the user before acting when a task is ambiguous.

What this means

Google is staging Carbon's release by audience, starting with cyber defenders, and has tied the rollout to guardrail iteration and government pre-release testing. Developers cannot yet use the model, so the pricing is a stated launch figure rather than something available to buy today.

The benchmark claims are mostly Google's own reporting, and the internal safety and secure-code figures have not been independently verified. The Intelligence Index placement, the SEC-bench Pro, ExploitGym and CWE-bench scores, and the WebDev Arena rank are external benchmarks, but Google presents the results. Context window and parameter count were not part of the announcement.